Performance of multi-model ensembles for the simulation of temperature variability over Ontario, Canada

Abstract

Climate ensembles utilize outputs from multiple climate models to estimate future climate patterns. These multi-model ensembles generally outperform individual climate models. In this paper, the performance of seven global climate model and regional climate model combinations were evaluated for Ontario, Canada. Two multi-model ensembles were developed and tested, one based on the mean of the seven combinations and the other based on the median of the same seven models. The performance of the multi-model ensembles were evaluated on 12 meteorological stations, as well as for the entire domain of Ontario, using three temperature variables (average surface temperature, maximum surface temperature, and minimum surface temperature). Climate data for developing and validating the multi-model ensembles were collected from three major sources: the North American Coordinated Regional Downscaling Experiment, the Digital Archive of Canadian Climatological Data, and the Climactic Research Unit’s TS v4.00 dataset. The results showed that the climate ensemble based on the mean generally outperformed the one based on the median, as well as each of the individual models. Future predictions under the Representative Concentration Pathway 4.5 (RCP4.5) scenario were generated using the multi-model ensemble based on the mean. This study provides credible and useful information for climate change mitigation and adaption in Ontario.